Solving Dual Optimization Problems in Identification and Performance of Fed-Batch Bioreactors

Abstract In this paper some aspects of the duality found within the following two tightly interconnected problems are discussed: (i) (model based) process performance optimization, and (ii) information content optimization for model identification. In the two case studies presented, the volumetric substrate feed rate into a fedbatch bioreactor in which one biomass grows on one limiting substrate is the control input to be optimized. Unstructured kinetic models in which the specific growth rate is a function of the substrate concentration only are considered. In case study #1 the optimization of a cost criterion combining both a performance cost and a parameter identification cost is discussed. Case study #2 is entirely devoted to information content optimization. A novel combined input design criterion for parameter estimation is constructed. For specific values of the weighting factor, it reduces to the modified E-criterion, the E-criterion, or the D-criterion.